odf regex matches, 0.56.2

2 files changed, 17 insertions, 6 deletions

diff --git a/CHANGELOG b/CHANGELOG
index b7f3df00..d8d5e4ff 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -6,11 +6,20 @@ Reverse Chronological:
 
 %% STABLE MANIFEST
 
+%% sisu_0.56.2.orig.tar.gz (2007-07-31:31/2)
+http://www.jus.uio.no/sisu/pkg/src/sisu_0.56.2.orig.tar.gz
+  sisu_0.56.2.orig.tar.gz
+  sisu_0.56.2-1.dsc
+  sisu_0.56.2-1.diff.gz
+
+  * odf, a few regex matches, a new fix (url), and an old one (grouped text
+    with change to fontface)
+
 %% sisu_0.56.1.orig.tar.gz (2007-07-31:31/2)
 http://www.jus.uio.no/sisu/pkg/src/sisu_0.56.1.orig.tar.gz
-  sisu_0.56.1.orig.tar.gz
-  sisu_0.56.1-1.dsc
-  sisu_0.56.1-1.diff.gz
+  6740ed538a5558cead0228315b811d9f 1272026 sisu_0.56.1.orig.tar.gz
+  1270f327c735cca3ea077950e5229f25 606 sisu_0.56.1-1.dsc
+  87e12b0b18192346e4a98c3127a73968 142382 sisu_0.56.1-1.diff.gz
 
   * dal, endnote / url matching refinement
 
diff --git a/lib/sisu/v0/odf.rb b/lib/sisu/v0/odf.rb
index 381e23d4..f6d6d942 100644
--- a/lib/sisu/v0/odf.rb
+++ b/lib/sisu/v0/odf.rb
@@ -279,7 +279,7 @@ module SiSU_ODF
       end
       def text_link(para)
         para.gsub!(@serial,'')
-        m=para.scan(/(\{([^}]+?)\}((?:https?|ftp)\S+?))([;.,]?$)/) #sort
+        m=para.scan(/(\{([^}]+?)\}((?:https?|ftp)\S+?))([;.,]?(?=\s|$))/) #sort
         if m
           m.each do |i|
             txt,url,trail=i[1],i[2]
@@ -297,8 +297,8 @@ module SiSU_ODF
         para.gsub!(@serial,'')
         para.gsub!(/\b[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/,
           %{<text:a xlink:type="simple" xlink:href="\\1">\\1</text:a>\\2}) #http ftp matches escaped, no decoration
-        para.gsub!(/((?:^|\s)[}])((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/,
-          %{\\1<text:a xlink:type="simple" xlink:href="\\2">\\2</text:a>\\3}) #special case \{ e.g. \}http://url
+        para.gsub!(/((?:^|\s)[}])((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?=\s|$))/,
+          '\1<text:a xlink:type="simple" xlink:href="\2">\2</text:a>\3') #special case \{ e.g. \}http://url
         para.gsub!(/(^|\s)((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?=\s|$))/,
           %{\\1#{@url_brace.xml_open}<text:a xlink:type="simple" xlink:href="\\2">\\2</text:a>#{@url_brace.xml_close}\\3}) #http ftp matches with decoration
         #para.gsub!(/\b((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([;.,]?(?:\s|$))/, also works
@@ -346,6 +346,7 @@ module SiSU_ODF
       def group_clean(para)
         para.gsub!(/&amp;nbsp;|&nbsp;/,'&#160;')
         para.gsub!(/</,'&lt;'); para.gsub!(/>/,'&gt;')
+        para.gsub!(/&lt;(text:span text:style-name="T[1-5]"|\/text:span)&gt;/,'<\1>') #works, not ideal
         para.gsub!(/&lt;br(?:\s+\/)?&gt;/,'<br />')
         #para.gsub!(/\s\s/,'&#160;&#160;')
         para
@@ -382,6 +383,7 @@ module SiSU_ODF
           parray << %{<text:p text:style-name="P5">#{parablock}</text:p>} if parablock =~/\S+/
         end
         para=parray.join + '<text:p text:style-name="Standard"/>'
+        para
       end
       def table(para)                                                            #
         if para =~/<!Th?.+/ # tables come as single block